Method of Detecting an Object Using a Camera

1. A method of detecting an object input via a camera, the method comprising: dividing image data including the object into pre-established areas and creating a detection window for each area; calculating a histogram for each created detection window; determining, for each one of the detection windows, weights for each pixel located within the one of the detection windows according to a determined similarity between a pre-stored histogram corresponding to the object and the histogram corresponding to the one of the detection windows; moving each of the detection windows in a direction corresponding to the determined weights according to an average movement, such that the detection windows converge towards the object; and selecting, from among the detection windows, a detection window having a histogram with maximum similarity to the pre-stored histogram corresponding to the object from the detection windows converging to the object to thereby detect and track the location of the object.

2. The method of claim 1 , wherein each area is selected according to a difference between a previous histogram and a subsequent histogram in a detection window corresponding to the area according to movement of the object for each frames of the image data, and wherein a detection window that is closest to the object is detected from among the detection windows by comparing the detection window in a selected area to a characteristic value of each pixel in the pre-stored histogram corresponding to the object.

3. The method of claim 1 , wherein the pre-stored histogram of the object is defined by: T = { t n } n = 1 , … ⁢ , m , ∑ n = 1 m ⁢ t n = 1 , where n is an index of a bin for the pre-stored histogram T, m is a total number of bins in the pre-stored histogram T, and t n is a histogram of n.

4. The method of claim 1 , wherein the histogram is created in each of the detection windows for each of the image data frames, with the histogram defined by a detection window represented by: R i = { r n i } n = 1 , … ⁢ , m , ∑ n = 1 m ⁢ r n i = 1. ⁢ ⁢ for ⁢ ⁢ i = 1 , … ⁢ , k , where i represents an i th area in an image window to which the object is input, k represents a total number of the detection windows in which the histogram is to be calculated, and m is a total number of bins in the pre-stored histogram.

5. The method of claim 1 , further comprising: comparing, for consecutive frames, a similarity level of the histogram in each of the detection windows to a corresponding window in a different consecutive frame; and deleting any detection windows that have substantially identical pixel values in consecutive frames.

6. The method of claim 5 , wherein the similarity of each of the frames in the detection windows is determined according to f ⁡ ( R i t , R i t - 1 ) = ∑ n = 1 m ⁢ r n i , t * r n i , t - 1 .

7. The method of claim 1 , wherein each histogram is produced in a multi-dimensional space using edge information obtained through detecting edge and Red/Green/Blue (RGB) channels of each pixel.

8. The method of claim 1 , wherein the weights are determined according to w l i = ∑ n = 1 m ⁢ t n r n i ⁢ δ ⁡ [ b ⁡ ( x l ) - n ] , where w l i is a weight of an l th pixel in an i th area for which a weight is determined, b(x l ) is a function for obtaining an index of a histogram to which a color of a pixel at a location x l belongs, n is an index of a bin for the histogram T, and m is a total number of bins in the pre-stored histogram.

9. The method of claim 1 , wherein, in the convergence of the detection window towards the object, the average movement of each detection window in the direction corresponding to the determined weights is repeatedly performed in order to calculate a subsequent location of the detection window, according to while ⁢ ⁢  x new - x old  > ε x new ← ∑ x k old ⁢ w k i ∑ w k i , where w k i is a weight of a k th pixel in the an i th area for which a weight is determined, ε is a critical value, and x is a location of the detection window.

10. The method of claim 1 , wherein the similarity varies according to movement of the object in an image window.

11. The method of claim 1 , wherein the pre-stored histogram reflects a characteristic value of the object to be detected.

12. The method of claim 1 , wherein the pre-established areas are adjusted according to a size of a screen displaying the image data.

13. The method of claim 1 , wherein the average movement of each detection window is determined pixel by pixel in the image window to which the object is input.

14. A non-transitory computer-readable recording medium, in which a program is provided for performing a method of detecting an object using a camera, the method comprising: dividing image data including the object into pre-established areas and creating a detection window for each area; calculating a histogram for each created detection window; determining, for each one of the detection windows, weights for each pixel located within the one of the detection windows according to a determined similarity between a pre-stored histogram corresponding to the object and the histogram corresponding to the one of the detection windows; moving each of the detection windows in a direction corresponding to the determined weights according to an average movement, such that the detection windows converge towards the object; and selecting, from among the detection windows, a detection window having a histogram with maximum similarity to the pre-stored histogram corresponding to the object from the detection windows converging to the object to thereby detect and track the location of the object.

15. A portable terminal comprising a recording medium, wherein a program is provided for performing a method of detecting an object using a camera, wherein the method includes: dividing image data including the object into pre-established areas and creating a detection window for each area; calculating a histogram for each created detection window; determining, for each one of the detection windows, weights for each pixel located within the one of the detection windows according to a determined similarity between a pre-stored histogram corresponding to the object and the histogram corresponding to the one of the detection windows; moving each of the detection windows in a direction corresponding to the determined weights according to an average movement, such that the detection windows converge towards the object; and selecting, from among the detection windows, a detection window having a histogram with maximum similarity to the pre-stored histogram corresponding to the object from the detection windows converging to the object to thereby detect and track the location of the object.